Kazem Taram

A Harmony of Ice and Fire: Mitigating Cold Last-Level Cache Effects in Serverless Computing on Commodity Hardware

Compute eXpress Leak: New Cache Attacks Using CXL Devices

SnoopDog is presented, a forward-looking cross-host timing attack that distinguishes whether a cache line is privately owned by another host based on access latency, and end-to-end cross-host side-channel attacks, including recovery of tokens from large language model inference when model weights are shared across host...

Leaky Language Models: Stealing Architecture and Inference Optimizations via Per-Token Timing

This work presents LeakyLMs, a set of attacks that leak proprietary model, architecture, and deployment information from production language models. LeakyLMs is the first to demonstrate that key model and deployment details can be inferred using only token generation timing, even when interacting through remote APIs. L...